AI is psychologically dangerous. It's the greater risk. It can influence people do to things.
The environment talk is perhaps relevant, but if eventually AI becomes ecologically friendly, then people who are against it would be left with no argument.
I'm skipping over to the core of the problem right away. It's a risk to human minds, always. The eco talk is probably relevant, but most likely it's more resounding as a bait to make you look like a tree-hugger that doesn't know what you're talking about.
To an extent, yes, but companies are investing more into renewable sources of energy anyway, and you might even get a demand-side effect where the cost of supply of energy is incentivized to go down due to higher competition.
But I assume your friends are justifying this in more of an ideological way than you'd think, because there are lots of things that are more environmentally damaging than data centers, and even for data centers it's not like we haven't been using Google and Facebook for 20 years when the same concerns of energy would've been there.
What I've seen is people dressing up the Luddite argument of losing economics of living into a lot of other arguments like energy usage, when in reality I doubt they really give a shit about that over environmental degradation in other areas of their life.
RAM is the reason LLMs are so power inefficient. Shuttling weights and results from RAM to compute and back for everything is where most of the power goes.
It doesn't have to be that way. For a sufficiently large load, it makes sense to use reconfigurable hardware and bake in the constants and s dataflow at runtime.
Think of it like using an array of FPGAs large enough to hold the whole model unwound, yet that could be configured in seconds at runtime. You'd get tokens at 100 MHz or more .
You would think saving 95% or more on power and infrastructure for a given token rate would be worth it, especially when contemplating Trillion dollar outlays.
Of course you are wrong. "IEA's models project that data centres will use 945 terawatt-hours (TWh) in 2030, roughly equivalent to the current annual electricity consumption of Japan." [1] And don't forget the recent announcement by Fermi America to build a 6 gigawatt nuclear datacenter in Texas. And what are we getting in return:
(1) AI is primarily used, by and far, to accelerate consumerism.
(2) AI has very few applications that are actually solving the world's problems (the world's problems are mostly nontechnical problems). The cited ones of scientific or medical are either too abstract, or else they are likely to make the problem worse. And in the case of medical, maybe the AI applications will help a few hundreds of thousands of people – but hurt/kill many more with its contribution to climate change. So not worth it.
I consider those who promote AI to be enemies of humanity and biological life since they are using a commodity that we should be using less of with reckless abandon.
1. https://hardware.slashdot.org/story/25/04/10/2019233/data-ce...
>I keep getting told by friends not to use AI because it is destroying the environment.
I side with them. If you're environmentally conscious you absolutely cant support AI. The amount of energy and the backing of basically every AI is fossil fuels.
You should oppose AI use.
I don't personally think there's that much value to this argument. Compare, for instance the consumption of 1 hour of tv vs. 1 hour of GPT usage:
A single AI chat message can consume 0.34 watt-hours of energy (1). So, let's say a hundred messages in an hour (quite an aggressive session) would be 34 watt-hours of energy.
An LCD TV running for an hour consumes about 100 watt-hours of energy, depending on size, LED, vs. OLED etc. (2).
I think AI does help people do better research faster, which is a significant uplift to humanity, while I do not see anyone specifically curbing their TV usage. We should probably focus our effrots on helping people use AI better and meanwhile build more nuclear energy plants, imo.
(1): https://epoch.ai/gradient-updates/how-much-energy-does-chatg...
(2): https://santannaenergyservices.com/how-many-watts-does-a-tv-...
---
And then consider the amount of energy traditionally required by one human to do the same research tasks. Also quite significant.
I think we should be focused on making the more efficient, for sure! But I don't buy that the arguments based on energy consumption are very strong.